SwinGALE: fusion of swin transformer and attention mechanism for GAN-augmented liver tumor classification with enhanced deep learning

Liver diseases represent a significant challenge to global healthcare systems, necessitating accurate and timely diagnosis for effective intervention. However, the intricate nature of liver tumor multi-classification remains a daunting obstacle. In this work, we provide a novel framework that integrates state-of-the-art technologies, including Generative Adversarial Networks (GANs), Convolutional Block Attention Module (CBAM), and Enhanced Channel Attention (ECA), within a deep learning architecture. Leveraging the comprehensive Duke Liver dataset, our approach synthesizes GAN-generated data to augment the training dataset and employs attention mechanisms to discern crucial details within medical images. Our ensemble model, incorporating CBAM with VGG19, achieves a remarkable accuracy of 99.29% in liver tumor classification. This research heralds a significant advancement in liver disease diagnosis, offering a promising avenue to improve patient outcomes.